General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: May 12, 2021

A modular metabolic engineering approach for the production of 1,2-propanediol fromglycerol by Saccharomyces cerevisiae

Islam, Zia-Ul; Klein, Mathias; Aßkamp, Maximilian R.; Ødum, Anders Sebastian Rosenkrans; Nevoigt,Elke

Published in:Metabolic Engineering

Link to article, DOI:10.1016/j.ymben.2017.10.002

Publication date:2017

Document VersionPeer reviewed version

Link back to DTU Orbit

Citation (APA):Islam, Z-U., Klein, M., Aßkamp, M. R., Ødum, A. S. R., & Nevoigt, E. (2017). A modular metabolic engineeringapproach for the production of 1,2-propanediol from glycerol by Saccharomyces cerevisiae. MetabolicEngineering, 44, 223-235. https://doi.org/10.1016/j.ymben.2017.10.002

1

A modular metabolic engineering approach for the production of

1,2-propanediol from glycerol by Saccharomyces cerevisiae

Zia-ul Islam1†, Mathias Klein1†, Maximilian R. Aßkamp1, Anders S.R. Ødum2, Elke Nevoigt1*

1 Department of Life Sciences and Chemistry, Jacobs University Bremen gGmbH, Campus Ring 1,

28759 Bremen, Germany

2 Department of Systems Biology, Building 223, Søltofts Plads, Technical University of Denmark,

Lyngby 2800 Kgs, Denmark

† Equal contributors

* Corresponding author: Mailing address: Jacobs University Bremen gGmbH, Department of Life

Sciences and Chemistry, Campus Ring 1, 28759 Bremen, Germany; Phone: +49-421-2003541; Fax:

+49-421-2003249; E-mail: e.nevoigt@jacobs-university.de

KEYWORDS

Yeast, Saccharomyces cerevisiae, glycerol, 1,2-propanediol, dihydroxyacetone, triosephosphate

isomerase

RUNNING TITLE

1,2-propanediol production from glycerol in Saccharomyces cerevisiae

2

ABSTRACT

Compared to sugars, a major advantage of using glycerol as a feedstock for industrial bioprocesses is

the fact that this molecule is more reduced than sugars. A compound whose biotechnological

production might greatly profit from the substrate’s higher reducing power is 1,2-propanediol (1,2-

PDO). Here we present a novel metabolic engineering approach to produce 1,2-PDO from glycerol in

S. cerevisiae. Apart from implementing the heterologous methylglyoxal (MG) pathway for 1,2-PDO

formation from dihydroxyacetone phosphate (DHAP) and expressing a heterologous glycerol

facilitator, the employed genetic modifications included the replacement of the native FAD-

dependent glycerol catabolic pathway by the `DHA pathway´ for delivery of cytosolic NADH and the

reduction of triosephosphate isomerase (TPI) activity for increased precursor (DHAP) supply. The

choice of the medium had a crucial impact on both the strength of the metabolic switch towards

fermentation in general (as indicated by the production of ethanol and 1,2-PDO) and on the ratio at

which these two fermentation products were formed. For example, virtually no 1,2-PDO but only

ethanol was formed in synthetic glycerol medium with urea as the nitrogen source. When nutrient-

limited complex YG medium was used, significant amounts of 1,2-PDO were formed and it became

obvious that the concerted supply of NADH and DHAP are essential for boosting 1,2-PDO production.

Additionally, optimizing the flux into the MG pathway improved 1,2-PDO formation at the expense of

ethanol. Cultivation of the best-performing strain in YG medium and a controlled bioreactor set-up

resulted in a maximum titer of > 4 g L-1 1,2-PDO which, to the best of our knowledge, has been the

highest titer of 1,2-PDO obtained in yeast so far. Surprisingly, significant 1,2-PDO production was also

obtained in synthetic glycerol medium after changing the nitrogen source towards ammonium

sulfate and adding a buffer.

3

1. INTRODUCTION

A major challenge for the production of chemicals by industrial biotechnology is the fact that the

commonly used sugar-based substrates inevitably result in low carbon yields which can be attributed

to the high oxidation state (O:C ratio) of sugars relative to many products (Burk and Van Dien, 2016).

Although glycerol is still less reduced than hydrocarbons derived from fossil resources, the degree of

reduction (4.667) is higher than that of common sugars such as glucose (4.000) and xylose (4.000)

(Yazdani and Gonzalez, 2008). Therefore, the utilization of glycerol as a substrate or co-substrate in

the fermentative production of chemicals can lead to higher maximum theoretical product yields of

compounds with a relatively high degree of reduction as well as of molecules whose biochemical

production from sugars is not redox-neutral but requires additional reducing equivalents such as 1,2-

propandiol or succinic acid (Murarka et al., 2008; Yazdani and Gonzalez, 2007; Zhang et al., 2013).

Glycerol is an unavoidable by-product of the biodiesel industry since transesterification of vegetable

oils or animal fats generates app. 10% (w/w) of glycerol (Karinen and Krause, 2006). As a result of the

growing demand for renewable transportation fuels, the production of biodiesel steadily increased

during the last decades accompanied by the accumulation of huge amounts of crude glycerol

(Mattam et al., 2013). In principle, this crude glycerol has to be considered as a waste stream since it

still contains various residual components from the transesterification process such as methanol,

NaOH, carry-over fats/oils as well as low amounts of proteins, minerals and sulfur compounds

(Thompson and He, 2006). The utilization of excess crude glycerol as a feedstock for the microbial

production of various value-added chemicals in industrial bioprocesses has been considered as a

promising approach for its valorization (da Silva et al., 2009; Johnson and Taconi, 2007; Yang et al.,

2012).

The yeast Saccharomyces cerevisiae has become a favorite platform organism for the production of

various chemical compounds by means of metabolic engineering. In fact, re-routing the carbon flux in

this organism allowed the commercial production of advanced biofuels (e.g. isoprenoids, isobutanol),

4

bulk chemicals (e.g. succinic acid) as well as pharmaceuticals and neutraceuticals (Borodina and

Nielsen, 2014; Hong and Nielsen, 2012; Nevoigt, 2008). Major factors driving the increasing

metabolic engineering incentives in S. cerevisiae are its robustness in industrial setups and its easy

accessibility for genetic engineering including the tremendous developments with regard to DNA

assembly, chromosomal engineering and genome editing as comprehensively reviewed by Chao et al.

(2015) and David and Siewers (2015).

Most of the commonly employed industrial and laboratory strains of S. cerevisiae utilize glycerol only

in the presence of complex medium supplements such as amino acids and nucleic bases but do not

grow in synthetic medium containing glycerol as the sole carbon source (Merico et al., 2011; Swinnen

et al., 2013). Therefore, glycerol has been rather neglected as a feedstock for S. cerevisiae-based

bioprocesses. Still, certain wild-type S. cerevisiae isolates including the strain CBS 6412 are able to

moderately grow in synthetic glycerol medium and a haploid segregant of this strain (CBS 6412-13A)

showed a maximum specific growth rate (µmax) of ~0.13 h-1 (Swinnen et al., 2013). The genetic

determinants underlying the latter segregant’s positive phenotype with regard to growth on glycerol

in comparison to the non-growing popular laboratory strain CEN.PK113-1A have been unraveled

recently (Swinnen et al., 2016). Moreover, the maximum specific growth rate of CBS 6412-13A in

synthetic glycerol medium has been even further improved up to ~0.18 h-1 by the expression of a

glycerol facilitator (Klein et al., 2016b). Several facilitators originating from various yeast species that

naturally show superior growth in glycerol medium as compared to S. cerevisiae (such as

Cyberlindnera jadinii, Yarrowia lipolytica, Pachysolen tannophilus and Komagataella phaffii syn.

Pichia pastoris) had a similar positive effect on glycerol utilization.

Wild-type S. cerevisiae strains exclusively utilize the so-called G3P (L-glycerol-3-phosphate) pathway

for catabolizing glycerol as recently reviewed by Klein et al. (2016c). This pathway transfers the

electrons derived from G3P oxidation directly via FADH2 to the cell’s mitochondrial respiratory chain

thereby ‘squandering’ the substrate’s aforementioned superior reducing power. In fact, the

respective electrons have to be available in the form of cytosolic NADH in order to support the

5

formation of reduced fermentation products. Recently, we replaced the endogenous G3P glycerol

catabolic pathway by the so-called DHA pathway which transfers the electrons to cytosolic NAD+

thereby delivering NADH for fermentative pathways (Klein et al., 2016a; Nevoigt, 2017). S. cerevisiae

strains engineered in this way show similar or even better glycerol utilization as compared to the

corresponding wild-type strains offering the possibility to employ the pathway replacement

approach for the production of reduced chemicals from glycerol in S. cerevisiae.

1,2-propanediol (1,2-PDO, propylene glycol) is one example for a reduced compound (degree of

reduction per Cmol: 5.333), whose biotechnological production may greatly profit from using glycerol

as a feedstock. 1,2-PDO is an important commodity chemical applied in food, drug, and cosmetic

industries. It is frequently used for the preparation of unsaturated polyester resins, for the

production of biodegradable plastics, as a solvent in photographic industries, as an aircraft de-icing

fluid and due to its lower toxicity as an environmentally friendly substitute for ethylene glycol in

automotive antifreeze (Berrios-Rivera et al., 2003; Niimi et al., 2011; Shelley, 2007). A wide range of

bacteria is naturally able to produce 1,2-PDO by fermentation, such as Thermoanaerobacterium

thermosaccharolyticum (Cameron and Cooney, 1986), Clostridium sphenoides (Tran-Din and

Gottschalk, 1985), Bacteroides ruminocola (Turner and Roberton, 1979), Klebsiella pneumoniae

(Badia et al., 1985) and E. coli (Boronat and Aguilar, 1979; Boronat and Aguilar, 1981; Gonzalez et al.,

2008; Murarka et al., 2008). Interestingly, there has also been a very early report about some aerobic

1,2-PDO production by several non-conventional yeast species (Suzuki and Onishi, 1968) and small

amounts have been detected in several industrial S. cerevisiae strains (Dowd et al., 1994). However,

one has to note that a full metabolic pathway for the production of 1,2-PDO has never been

identified in yeast. In addition, the respective cultivations were performed in undefined media which

may contain one or several of the pathway intermediates. These intermediates could have been

converted to 1,2-PDO by endogenous, non-specific oxidoreductase activities (Bennett and San,

2001).

6

Two main natural biochemical pathways leading to 1,2-PDO production have been discovered in

bacteria. A widely spread pathway, which exists for example in E. coli and Salmonella typhimurium,

depends on the fermentation of the deoxy hexose sugars L-rhamnose and L-fucose (Badia et al.,

1985; Bennett and San, 2001; Boronat and Aguilar, 1981). A second natural route identified in

T. thermosaccharolyticum and C. sphenoides converts the glycolytic carbon intermediate

dihydroxyacetone phosphate (DHAP) by a methylglyoxal synthase (MGS) to methylglyoxal (MG)

followed by two subsequent NAD(P)H dependent reductions catalyzed by aldo-keto reductases

(AKRs) to 1,2-PDO (see Figure 1) (Bennett and San, 2001; Cameron and Cooney, 1986; Tran-Din and

Gottschalk, 1985). 1,2-PDO formation by wild-type E. coli had been originally assumed to be

restricted to the fermentation of the deoxy hexose sugars as mentioned above. However, more

recent studies demonstrated that the so-called ʽmethylglyoxal pathwayʼ is also used in E. coli to

anaerobically ferment glycerol where 1,2-PDO production is used as a redox sink in order to cope

with the surplus cytosolic NADH derived from biomass formation (Gonzalez et al., 2008; Murarka et

al., 2008). As the latter pathway allows for the production of 1,2-PDO from inexpensive sugars or

glycerol it has been the preferred route in the context of rational engineering of microorganisms for

1,2-PDO production. So far, the bacterium E. coli has been the mainly employed chassis in numerous

engineering studies aiming at the fermentative production of 1,2-PDO using either glucose (Altaras

and Cameron, 1999; Altaras and Cameron, 2000; Berrios-Rivera et al., 2003; Cameron et al., 1998;

Huang et al., 1999; Jain et al., 2015a; Jain et al., 2015b; Lee et al., 2016) or glycerol (Clomburg and

Gonzalez, 2011) as the sole carbon source. The ʽmethylglyoxal pathwayʼ was also used to produce

1,2-PDO from glucose in engineered cells of Corynebacterium glutamicum (Niimi et al., 2011; Siebert

and Wendisch, 2015) and from CO2 by metabolic engineering of the cyanobacterium Synechococcus

elongatus (Li and Liao, 2013). Eukaryotic microorganisms previously engineered for 1,2-PDO

production include S. cerevisiae (Jeon et al., 2009; Jung et al., 2008; Jung et al., 2011; Lee and DaSilva,

2006) and K. phaffii (P. pastoris) (Barbier et al., 2011). Notably, the highest titers of 1,2-PDO such as

7

5.62 g/L for E. coli (Clomburg and Gonzalez, 2011) and 2.19 g/L for S. cerevisiae (Jung et al., 2011)

were achieved when using glycerol as the carbon source.

Here, we present a novel metabolic engineering study for the production of 1,2-PDO from glycerol in

the yeast S. cerevisiae. The current work was initiated since we assumed that our previously reported

glycerol catabolic pathway replacement (Klein et al., 2016a) might be crucial for providing sufficient

amounts of cytosolic NADH for the production of reduced chemicals such as 1,2-PDO. Apart from

obtaining the highest 1,2-PDO titer produced in yeast so far, we also got surprising new insights into

the physiology of glycerol metabolism in our engineered strains including unexpected effects caused

by the medium composition.

8

Figure 1 (2 column fitting image). Metabolic engineering for 1,2-propanediol (1,2-PDO) production from

glycerol in S. cerevisiae. Module `PDO´: `methylglyoxal pathway´ converting dihydroxyacetone phosphate

(DHAP) via methylglyoxal to 1,2-PDO by expressing E. coli methylglyoxal synthase (encoded by EcmgsA), E. coli

9

aldehyde reductase (encoded by EcyqhD) and E. coli glycerol dehydrogenase (encoded by EcgldA). Module

`DHA´: replacement of the native FAD-dependent glycerol catabolic pathway via L-glycerol-3-phosphate by the

NAD+-dependent `DHA pathway´ by deletion of the gene encoding glycerol kinase (GUT1), expression of a

glycerol dehydrogenase from O. parapolymorpha (Opgdh) and overexpression of endogenous

dihydroxyacetone kinase (encoded by DAK1). Module `TPI1down´: downregulation of the expression of TPI1

encoding triosephosphate isomerase by replacement of the gene’s native promoter by the weak TEF1

promoter variant TEFmut2 (Nevoigt et al., 2006). All constructed S. cerevisiae strains express the glycerol

facilitator from C. jadinii encoded by CjFPS1 in addition to the endogenous glycerol/H+ symporter encoded by

STL1. The latter modification has been considered to be useful as it resulted in a significantly improved glycerol

uptake in the genetic background of CBS 6412-13A (Klein et al., 2016b). Bold grey arrows highlight the carbon

flux from glycerol to 1,2-PDO as it is assumed based on the published knowledge regarding the in vivo and/or in

vitro activities of transport and enzymes as explained in the main text. Still, it has to be considered that the

heterologous enzymes encoded by EcgldA, EcyqhD and Opgdh have relatively broad substrate specificities and

might also contribute to other reduction/oxidation reactions in addition those allocated to the gene names in

this Figure. Moreover, unspecific/uncharacterized endogenous enzyme activities in S. cerevisiae might also

contribute to the different potential reactions from methylglyoxal to 1,2-PDO.

2. MATERIALS AND METHODS

2.1 Strains and their maintenance

All S. cerevisiae strains used in this study were based on CBS 6412-13A (Swinnen et al., 2013) and are

listed in Table 1. The yeast strains were routinely grown on solid YPD medium containing 10 g L-1

yeast extract, 20 g L-1 peptone, 20 g L-1 glucose, and 15 g L-1 agar. All strains, whose native TPI1

promoter had been replaced by the engineered TEF1 promoter variant TEFmut2, were grown on

solid YP glycerol medium (YPG) containing 30 g L-1 glycerol as the carbon source. If required, solid

medium was supplemented with G418 (200 mg L-1), phleomycin (20 mg L-1), nourseothricin

(100 mg L-1) or hygromycin B (300 mg L-1). E. coli DH5α was used for plasmid construction and

isolation, and cells were routinely grown in lysogeny broth (LB) (10 g L-1 peptone, 5 g L-1 yeast extract

and 10 g L-1 NaCl, pH 7.0). For selection and maintenance of plasmid containing cells, 100 mg L-1

ampicillin was added.

2.2 General molecular biology techniques

10

All primers used within the present study are listed in Table S1 (supplementary information).

Preparative PCRs for cloning were performed using Phusion® High-Fidelity DNA Polymerase (New

England BioLabs, Frankfurt am Main, Germany) or AccuPrime Pfx DNA polymerase (Thermo Fisher

Scientific, Waltham, MA, USA). PCR conditions were adapted to the guidelines of the respective

manufacturer. Restriction enzymes, FastAP alkaline phosphatase and T4 DNA ligase were obtained

from Thermo Fisher Scientific and used according to the manufacturer’s instructions. PCR products

were purified by using the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and DNA

fragments obtained after restriction were excised and purified using the QIAquick Gel Purification Kit

(Qiagen). Plasmids were isolated from E. coli DH5α overnight cultures using the QIAprep Spin

Miniprep Kit (Qiagen). Transformation of S. cerevisiae with plasmids as well as with linear DNA

fragments for genomic integration was performed according to the lithium acetate method

described by Gietz et al. (1995).

Table 1: Strains used in this study.

Strain / Modules Genome modifications Plasmid Source or reference

CBS 6412-13A - - Swinnen et al. (2013)

CBS 6412-13A CjFPS1 YGLCτ3:: TEF1p-CjFPS1-CYC1t - Klein et al. (2016b)

CBS 6412-13A DAK1OE gut1Δ

YPRCΔ15::ACT1p-DAK1-TPS1t; gut1::loxP-kanMX4-loxP

- Klein et al. (2016a)

FPS YGLCτ3:: TEF1p-CjFPS1-CYC1t p41ble-E-E This study

PDO-FPS YGLCτ3::TEF1p-EcmgsA-CYC1t:TDH3p-EcgldA-IDP1t-PGK1p-CjFPS1-RPL15At

p41ble-E-yqhD (ACT1p-EcyqhD-TPS1t)

This study

PDO-FPS-DHA YGLCτ3::TEF1p-EcmgsA-CYC1t:TDH3p-EcgldA-IDP1t-PGK1p-CjFPS1-RPL15At; YPRCΔ15::ACT1p-DAK1-TPS1t; gut1::loxP-kanMX4-loxP

p41ble-gdh-yqhD (TEF1p-Opgdh-CYC1t; ACT1p-EcyqhD-TPS1t)

This study

PDO-FPS-TPI1down YGLCτ3::TEF1p-EcmgsA-CYC1t:TDH3p-EcgldA-IDP1t-PGK1p-CjFPS1-RPL15At; tpi1p::TEFmut2p-TPI1

p41ble-E-yqhD (ACT1p-EcyqhD-TPS1t)

This study

PDO-FPS-TPI1down-DHA YGLCτ3::TEF1p-EcmgsA-CYC1t:TDH3p-EcgldA-IDP1t-PGK1p-CjFPS1-RPL15At; tpi1p::TEFmut2p-TPI1; YPRCΔ15::ACT1p-DAK1-TPS1t; gut1::loxP-kanMX4-loxP

p41ble-gdh-yqhD (TEF1p-Opgdh-CYC1t; ACT1p-EcyqhD-TPS1t)

This study

PDO-FPS-TPI1down-DHA EcmgsA2

YGLCτ3::TEF1p-EcmgsA-CYC1t:TDH3p-EcgldA-IDP1t-PGK1p-CjFPS1-RPL15At; tpi1p::TEFmut2p-TPI1; YPRCΔ15::ACT1p-DAK1-TPS1t; gut1::loxP-kanMX4-loxP; YPRCτ3::TDH3p-EcmgsA-IDP1t: TEF1p-Opgdh-CYC1t; YORWΔ22:: ACT1p-EcyqhD-TPS1t

- This study

11

2.3 Construction of plasmids and expression cassettes

All constructed plasmids are listed in Table S2 (supplementary information). Codon-optimized coding

sequences for E. coli mgsA, gldA and yqhD were obtained via GeneArtTM gene synthesis from Thermo

Fisher Scientific (Waltham, MA, USA). The plasmids for either the sole expression of E. coli yqhD

(EcyqhD) under the control of the ACT1 promoter and the TPS1 terminator or its expression in

combination with the O. parapolymorpha gdh (Opgdh), under the control of the TEF1 promoter and

the CYC1 terminator, were obtained by using Gibson isothermal assembly (Gibson et al., 2009). The

assembly was performed in the previously constructed S. cerevisiae expression vectors p41bleTEF or

p41bleTEF-Opgdh (Klein et al., 2016a). The ACT1 promoter and the TPS1 terminator sequences were

amplified from genomic DNA isolated from the S. cerevisiae strain S288c using primers 531 and 532,

and 533 and 534, respectively (see Table S1). The coding sequence of E. coli yqhD was amplified from

pMA-T-yqhD (Table S2) using primers 535 and 536, and the template vector was subsequently

removed by digestion with DpnI. The assembly reactions contained 15 μL of the reagent-enzyme mix,

0.05 pmol of KpnI-linearized and dephosphorylated p41bleTEF or p41bleTEF-Opgdh and 3-fold molar

excess of the inserts (promoter, coding sequence and terminator each 0.15 pmol) in a final volume of

20 µL. Reaction mixtures were incubated at 50°C for 1 h. E. coli DH5α cells were transformed with

5 μL of the reaction and the resulting vectors named p41ble-E-yqhD and p41ble-gdh-yqhD where “E”

specifies an empty expression cassette consisting solely of the TEF1 promoter and CYC1 terminator

sequences (Table S2). The empty vector control (carrying two empty expression cassettes) was

constructed accordingly by assembling only the ACT1 promoter and the TPS1 terminator in

p41bleTEF. The primers used for amplification were 531 and 538, and 533 and 537, respectively. The

resulting ‘empty’ vector was named p41ble-E-E and used for the construction of control strains.

The expression cassettes for C. jadinii FPS1 (CjFPS1) and S. cerevisiae DAK1 (ScDAK1) for genomic

integration were obtained by PCR amplification from the plasmids pUC18-CjFPS1 and pUC18-DAK1

(Table S2). Construction of these plasmids is described in Klein et al. (2016a). One of the two

expression cassettes constructed for integrating E. coli mgsA into the genome (EcmgsA under the

12

control of the S. cerevisiae TEF1 promoter and the CYC1 terminator) was obtained by cloning the

respective coding sequence into the multiple cloning site of p416TEF-KanMX (Table S2). For this

purpose the EcmgsA coding sequence was amplified from pMA-T-mgs (Table S2) using primers 276

and 277. The cloning via BamHI and EcoRI yielded the plasmid p416TEF-mgsA-KanMX. The second

EcmgsA expression cassette (EcmgsA2 under the control of the TDH3 promoter and the IDP1

terminator) and the cassette for expression of E. coli gldA (EcgldA) (also under the control of the

TDH3 promoter and the IDP1 terminator) were both assembled in pUC18 using Gibson isothermal

assembly (Gibson et al., 2009). Promoters and terminators were amplified from genomic DNA

isolated from the S. cerevisiae strain S288c using primer combinations 413/419 and 415/328 for the

expression cassette of EcmgsA2, and 324/323 and 327/328 for the expression cassette of EcgldA. The

coding sequences for EcmgsA and EcgldA were amplified from the plasmids pMA-T-mgs and pGA-

gldA using primer combinations 420/414 and 325/326, respectively. Both template vectors were

removed after PCR by digestion with DpnI. The target vector pUC18 was linearized using BamHI and

subsequently dephosphorylated to avoid self-ligation. The further steps for assembly and

transformation were the same as described above and the resulting plasmids named pUC18-gldA and

pUC18-mgsA. All constructed plasmids were verified by Sanger sequencing (GATC Biotech, Konstanz,

Germany).

2.4 Plasmids for CRISPR-Cas9 mediated genome editing in S. cerevisiae

For CRISPR-Cas9 mediated genome editing, the plasmids p414-TEF1p-Cas9-CYC1t-nat1 and p426-

SNR52p-gRNA.CAN1.Y-SUP4t-hphMX (Table S2) were used. Construction of the respective plasmids

has been reported by Klein et al. (2016a). The specific 20 nt sequence at the 5’-end of the gRNA

targeting the CAN1 gene was exchanged in p426-SNR52p-gRNA.CAN1.Y-SUP4t-hphMX by a 20 nt

sequence targeting the YGLCτ3 region on chromosome VII (Flagfeldt et al., 2009) by generating two

overlapping PCR products of the gRNA expression cassette and their subsequent assembly in the

same vector backbone according to Gibson et al. (2009). The target sequence within the YGLCτ3

locus was selected after analysis of the respective genomic region using the CRISPRdirect online tool

13

developed by Naito et al. (2015). The two overlapping fragments of the gRNA expression cassette

were obtained by PCR amplification from p426-SNR52p-gRNA.CAN1.Y-SUP4t (Table S2) using primers

460 and 464, and 465 and 463, respectively (Table S1), where the 5’-termini of primers 464 and 465

generated flanking sequences at the ends of the two PCR products which overlapped with the

YGLCτ3 target sequence for genomic integration. The primers 460 and 463 contained 5’-extensions

resulting in overlaps with PvuII-linearized p426-SNR52p-gRNA.CAN1.Y-SUP4t-hphMX. The accordingly

digested vector was gel-purified and Gibson isothermal assembly was performed as described above

(section 2.3) yielding p426-SNR52p-gRNA.YGLCτ3-SUP4t-hphMX.

2.5 S. cerevisiae strain construction

A detailed scheme for the construction of engineered CBS 6412-13A derivatives (Table 1) is shown in

Figure S1 (supplementary information). As the expression of EcyqhD and Opgdh in all constructed

strains except PDO-FPS-TPI1down-DHA EcmgsA2 was plasmid-based, the final step for completion was

always a transformation with p41ble-E-yqhD, p41ble-gdh-yqhD or p41ble-E-E.

Strains PDO-FPS and FPS

As mentioned in section 2.4 the long terminal repeat (LTR) YGLCτ3 on chromosome VII (Flagfeldt et

al., 2009) was selected as a target site for the genomic integration of the expression cassettes for

EcmgsA (under the control of the TEF1 promoter and the CYC1 terminator) and EcgldA (under the

control of the TDH3 promoter and the IDP1 terminator) in CBS 6412-13A. The seamless integration

was achieved in two steps. First, a counter-selectable, galactose-inducible growth inhibitory

sequence (referred to as GALp-GIN11M86) and the kanMX4 selectable marker were assembled and

integrated at the target site by homologous recombination. The GALp-GIN11M86 cassette was

amplified from plasmid pGG119 (Akada et al., 2002) using primers 175 and 155. The kanMX4

expression cassette was amplified from plasmid pUG6 (Gueldener et al., 1996) using primers 171 and

176. The PCR primers were modified at their 5’-ends generating homologous regions of ~40-60 bp at

the ends of the PCR products for assembly and integration at the target locus. In the second step, the

14

two aforementioned expression cassettes EcmgsA and EcgldA were assembled and integrated by

homologous recombination at the target locus replacing the entire GALp-GIN11M86/kanMX4

cassette. The respective expression cassettes were amplified from the previously constructed

plasmids p416TEFmgsA-KanMX (primers 257 and 302) and pUC18-gldA (primers 303 and 297),

respectively. Primer pairs were designed to generate PCR products whose flanking sequences were

homologous to the chromosomal regions upstream and downstream of the previously integrated

GALp-GIN11M86/kanMX4 cassette. After transformation, cells were plated on solid synthetic

medium containing 2% (w/v) galactose as the sole carbon source in order to induce the growth

inhibitory sequence GIN11M86 for counter-selection. The resulting intermediate strain was named

PDOmin (Figure S1). Subsequently, an expression cassette for CjFPS1 (under the control of the PGK1

promoter and the RPL15A terminator) was integrated together with a phleomycin resistance marker

(ble) downstream of EcmgsA and EcgldA. The CjFPS1 cassette was amplified with primers 390 and

392 from pUC18-CjFPS1, while the ble cassette was amplified from pUG66 (Gueldener et al., 2002)

using primers 391 and 393. The 5’-termini of the employed primers were designed to allow assembly

and integration of the generated PCR products into the genome of strain PDOmin, immediately

downstream of the previously integrated expression cassettes for EcmgsA and EcgldA, and the

resulting strain was named PDOmin-FPS. The phleomycin resistance cassette was subsequently

removed from this intermediate strain by using Cre-loxP-mediated recombination according to

Steensma and Ter Linde (2001) using the Cre-recombinase expressing plasmid pNatCre (Table S2).

Finally the resulting strain was transformed with p41ble-E-yqhD to yield the strain PDO-FPS. In order

to generate a control strain without an implemented pathway for 1,2-PDO production the previously

constructed strain CBS 6412-13A CjFPS1 (Klein et al., 2016b) was transformed with the ‘empty’ vector

p41ble-E-E (section 2.3; Table S2).

Strain PDO-FPS-DHA

In order to combine the modules for the 1,2-PDO pathway and the glycerol catabolic pathway

replacement (DHA pathway replacing the endogenous G3P pathway) in an FPS1 expressing strain of

15

CBS 6412-13A, the previously generated strain CBS 6412-13A DAK1OE gut1Δ was used as a basis

(Table 1) (Klein et al., 2016a). First, the expression cassettes for EcmgsA, EcgldA and CjFPS1 were

integrated into the YGLCτ3 region of this strain by employing the CRISPR-Cas9 system. For this

purpose, the expression of Cas9 in the strain CBS 6412-13A DAK1OE gut1Δ was achieved by

transformation with the plasmid p414-TEF1p-Cas9-CYC1t-nat1 (Table S2). The three expression

cassettes were PCR-amplified from the plasmids p416TEFmgsA-KanMX, pUC18-gldA and pUC18-

CjFPS1 (see section 2.3 and Table S2) using primers pairs 257/302, 303/552, and 550/551,

respectively. These primers contained 5’-extensions generating 40-60 bp sequences homologous to

regions directly upstream and downstream of the YGLCτ3 integration locus or to the respective

adjacent expression cassette (Figure S1). Co-transformation of the expression cassettes and plasmid

p426-SNR52p-gRNA.YGLCτ3-SUP4t-hphMX into the S. cerevisiae strain expressing the Cas9

endonuclease resulted in assembly and integration of all three expression cassettes at the target

locus. Positive transformants were selected on YPD agar containing both nourseothricin and

hygromycin B as described in section 2.1. Both plasmids were subsequently removed from the

resulting clone by serial transfers in YPD medium lacking the respective antibiotics. In order to

complete both modules (1,2-‘PDO’ and ‘DHA’), the resulting intermediate strain was transformed

with p41ble-gdh-yqhD (see section 2.3) yielding strain PDO-FPS-DHA.

Strains PDO-FPS-TPI1down and PDO-FPS-TPI1down-DHA

In order to downregulate TPI1 expression, the gene’s native promoter was replaced by a promoter

replacement cassette containing the weak TEF1 promoter variant TEFmut2p (Nevoigt et al., 2006).

The replacement cassette consisting of the kanMX4 selectable marker and the TEFmut2 promoter

variant was amplified from the plasmid p416-TEFmut2-yECitrine (Table S2) using primers 426 and

427. The primers contained 5’-extensions of 60 bp homologous to sequences directly upstream and

downstream of the endogenous TPI1 promoter. Strain PDOmin-FPS (Figure S1) was transformed with

the amplified cassette leading to the replacement of the endogenous TPI1 promoter by the

engineered promoter variant TEFmut2. Transformants were selected on medium containing glycerol

16

as the carbon source. Both dominant markers (the ble cassette downstream of CjFPS1 and the

kanMX4 cassette upstream of TPI1) in the intermediate strain PDOmin-FPS-TPI1down were

subsequently rescued by Cre-loxP mediated recombination as described for strain PDOmin-FPS (see

above). The resulting strain was transformed with plasmid p41ble-E-yqhD yielding strain PDO-FPS-

TPI1down. The previous strain (PDOmin-FPS-TPI1down after marker removal) also served as the basis for

the construction of strain PDO-FPS-TPI1down-DHA (Figure S1). First, the cassette for ScDAK1

overexpression (ACT1 promoter and TPS1 terminator) was integrated via the two-step seamless

integration procedure employing the counter-selectable, galactose-inducible growth inhibitory

sequence GALp-GIN11M86 as described above. The LTR YPRCΔ15 on chromosome XVI (Flagfeldt et

al., 2009) served as a target site and the GALp-GIN11M86 growth inhibitory sequence (amplified

from plasmid pGG119 with primers 298 and 155) was first integrated at this position together with

the phleomycin resistance marker (amplified from plasmid pUG66 with primers 171 and 299)

(Table S2). The ScDAK1 expression cassette was amplified from plasmid pUC18-DAK1 using primers

388 and 389 and subsequently integrated at the target locus by homologous recombination replacing

the entire GALp-GIN11M86/ble cassette. Afterwards, the endogenous GUT1 coding sequence was

disrupted by integration of the kanMX4 cassette amplified from plasmid pUG6 using primers 111 and

112 yielding the intermediate strain PDOmin-FPS-TPI1down DAK1OE gut1Δ. Finally, the resulting

intermediate strain was transformed with plasmid p41ble-gdh-yqhD to finalize the construction of

the strain PDO-FPS-TPI1down-DHA.

Strain PDO-FPS-TPI1down-DHA EcmgsA2

In order to implement a second expression cassette for mgsA into the strain and, at the same time,

obtain a CBS 6412-13A derivative that carries all above-described genetic modifications for 1,2-PDO

production in the genome, the strain PDO-FPS-TPI1down-DHA EcmgsA2 was generated (Figure S1). The

aforementioned intermediate strain PDOmin-FPS-TPI1down DAK1OE gut1Δ was used as a starting point

for the integration of the second expression cassette for EcmgsA (EcmgsA2, coding sequence under

the control of the TDH3 promoter and the IDP1 terminator). Similarly to the expression cassettes of

17

EcmgsA, EcgldA and ScDAK1, the cassette was integrated together with that of Opgdh into the

genome via the two-step seamless integration procedure described above using the LTR YPRCτ3 on

chromosome XVI as a target site (Flagfeldt et al., 2009). First, the GALp-GIN11M86 growth inhibitory

sequence (amplified from plasmid pGG119 with primers 300 and 155) and the phleomycin resistance

marker (amplified from plasmid pUG66 with primers 171 and 301) were assembled and integrated at

the target locus by homologous recombination. The whole Galp-GIN11M86/ble cassette was

subsequently replaced by assembly and integration of the EcmgsA2 and Opgdh expression cassettes

(amplified with primers 418 and 422 from plasmid pUC18-mgsA and with primers 416 and 421 from

plasmid p41bleTEF-Opgdh). Subsequently, the above-described expression cassette for E. coli yqhD

was integrated along with the phleomycin resistance marker at the YORWΔ22 locus (Flagfeldt et al.,

2009) yielding strain PDO-FPS-TPI1down-DHA EcmgsA2 (see Figure S1). The EcyqhD cassette was

amplified with primers 663 and 664 from plasmid p41ble-E-yqhD, while the ble cassette was

amplified from plasmid pUG66 (Gueldener et al., 2002) using primers 665 and 666. The 5’-termini of

the employed primers were designed to allow assembly and integration of the generated PCR

products at the target site in the genome.

2.6 Isolation of genomic DNA from S. cerevisiae transformants and diagnostic PCR

Correct integrations of all expression and disruption cassettes were verified by diagnostic PCR using

TaKaRa Ex Taq Polymerase and buffer according to the manufacturer’s guidelines (Merck KGaA,

Darmstadt, Germany). For this purpose, single colonies obtained after transformations were re-

streaked on respective agar plates. Approximately 50 mg of cells from these plates were suspended

in 200 μL of TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0). Subsequently, 300 mg of acid-washed glass

beads (diameter of 0.425 to 0.6 mm) and 200 μL of phenol:chloroform:isoamyl alcohol (25:24:1)

were added. The tubes were then vortexed at maximum speed for 2 min and centrifuged at 15,700 g

for 10 min according to a protocol modified from Hoffman and Winston (1987). The aqueous phase

(1 µL) was used as template in 20 µL PCR reactions. PCR primers were designed to bind upstream and

downstream of the genomic integration sites as well as within the integrated expression/deletion

18

cassette. For analyzing integrations of multiple expression cassettes, additional primers were

designed to produce amplicons covering the junctions between the individual integrated expression

cassettes. After each integration step, the presence of all previously integrated expression cassettes

was verified.

2.7 Media and cultivation conditions for the production of 1,2-PDO from glycerol

Media used for shake flask batch cultivations assessing 1,2-PDO production were synthetic or

complex media containing 60 mL L-1 glycerol as the sole carbon source. The synthetic medium

according to Verduyn et al. (1992) contained 3 g L-1 KH2PO4, 0.5 g L-1 MgSO4.7H2O, 15 mg L-1 EDTA,

4.5 mg L-1 ZnSO4.7H2O, 0.84 mg L-1 MnCl2.2H2O, 0.3 mg L-1 CoCl2.6H2O, 0.3 mg L-1 CuSO4.5H2O,

0.4 mg L-1 NaMoO4.2H2O, 4.5 mg L-1 CaCl2.2H2O, 3 mg L-1 FeSO4.7H2O, 1 mg L-1 H3BO3, and 0.1 mg L-1 KI.

In case ammonium sulfate was used as the nitrogen source 5 g L-1 (NH4)2SO4 was added. After heat

sterilization, an appropriate aliquot of vitamin stock solution was added. The final concentrations of

vitamins were 0.05 mg L-1 D-(+)-biotin, 1 mg L-1 D-pantothenic acid hemicalcium salt, 1 mg L-1 nicotinic

acid, 25 mg L-1 myo-inositol, 1 mg L-1 thiamine chloride hydrochloride, 1 mg L-1 pyridoxine

hydrochloride, and 0.2 mg L-1 4-aminobenzoic acid. In case urea was used as the nitrogen source, an

appropriate aliquot of a stock solution was added after heat sterilization to obtain a final

concentration of 2.8 g L-1. For buffering the synthetic medium, 100 mM of potassium hydrogen

phthalate was added to the medium prior autoclaving. The pH of the synthetic glycerol medium was

adjusted to 4.0 (for non-buffered medium) and to 5.0 (for medium buffered with potassium

hydrogen phthalate) with 2 M H3PO4. A pH of 4-5 was previously determined to be optimal for the

growth on glycerol of wild-type CBS 6412-13A (unpublished results). The complex media used were

YPG6% containing 10 g L-1 yeast extract, 20 g L-1 peptone and 60 mL L-1 glycerol or YG containing only

10 g L-1 yeast extract in addition to 60 mL L-1 glycerol. The pH of the complex media was adjusted to

6.0 with 2 M H3PO4. Except for the strain PDO-FPS-TPI1down-DHA-EcmgsA2, all liquid media were

supplemented with 30 mg L-1 phleomycin for plasmid maintenance.

19

For pre-cultivation, cells from a single colony were generally used to inoculate 3 mL of the respective

medium in a 10 mL glass tube and incubated at orbital shaking of 200 rpm and 30°C for 24 (complex

glycerol medium) to 48 h (synthetic glycerol medium). The pre-culture was used to inoculate 10 mL

of the same medium in a 100 mL Erlenmeyer flask adjusting an OD600 of 0.2. This intermediate culture

was cultivated at the same conditions until an OD600 of ~2.0 was reached. The appropriate culture

volume (in order to later adjust an OD600 of 0.2 in 50 mL of medium) was centrifuged at 800 g for

5 min and the supernatant discarded. The cell pellet was then resuspended in 50 mL of the

respective glycerol-containing medium in a 500 mL Erlenmeyer flask. The main cultures were

incubated at orbital shaking of 200 rpm and 30°C and samples for OD600 determination and HPLC

analysis were taken at regular time intervals.

2.8 Cultivations in controlled bioreactors

All reactor-based cultivations were carried out using Sartorious 1 L bioreactors (Sartorious, Stedim

Biotech, Göttingen, Germany) with equivalent working volumes and equipped with 2 Rushton six-

blade disc turbines. The pH electrode (Mettler Toledo, OH, USA) was calibrated according to

manufacturer’s standard procedures and pH was controlled by automatic addition of 2 M NaOH and

H2SO4. The bioreactor was sparged with sterile atmospheric air and the temperature was maintained

at 30°C throughout the cultivations. Culture conditions were: pH 5, stirring rate: 800 rpm and air

flow: 1 volume of air per volume of liquid per minute (vvm). Samples were taken across the

exponential growth phase to monitor growth (OD600) as well as for HPLC analysis for metabolite

profiling.

2.9 Analytical methods

Samples of culture supernatants were first filtered through 0.2 µm Minisart RC membrane filters

(Sartorius) and stored at -20°C until analysis. The concentrations of glycerol, 1,2-PDO and ethanol in

culture media were determined using a Waters HPLC system (Eschborn, Germany) equipped with

two consecutive Aminex HPX-87H cation-exchange columns (Biorad, München, Germany) using

20

5 mM H2SO4 as the mobile phase at a flow rate of 0.45 mL min-1 and a column temperature of 45°C.

Volumes of 20 μL of sample were used for injection. All components were detected using the Waters

2414 Refractive Index detector. Under these conditions, the retention times were 35.6 min for

glycerol, 45.4 min for 1,2-PDO and 57.7 min for ethanol. For quantification, standards were run in

appropriate concentrations (i.e. between 1 g L-1 and 100 g L-1 for glycerol, 0.075 g L-1 and 5 g L-1 for

1,2-PDO, and 0.394 g L-1 and 39.37 g L-1 for ethanol). The obtained raw data were processed using the

Breeze 2 software (Waters).

3. RESULTS

3.1 Modular metabolic engineering of S. cerevisiae for the production of 1,2-PDO from glycerol

All metabolic engineering steps in this study were conducted in the genetic background of our

previously described and intensively studied haploid meiotic segregant CBS 6412-13A originating

from the glycerol-utilizing wild-type S. cerevisiae strain CBS 6412 (Swinnen et al., 2016). This strain

exhibiting a µmax of 0.13 h-1 in synthetic glycerol medium was the best-performing glycerol-utilizing

S. cerevisiae strain available to us at the time when the current study has been initiated. As

demonstrated in our previous study, the strain CBS 6412-13A solely utilizes the native FAD-

dependent G3P pathway for glycerol catabolism (Swinnen et al., 2013).

As. S. cerevisiae does not produce 1,2-PDO naturally, the first engineering module implemented

during strain construction was the expression of the so-called `methylglyoxal pathway´ (see section 1

and Figure 1). The key enzyme of this pathway is methylglyoxal synthase (MGS) and several previous

studies have shown that even the sole expression of E. coli mgsA in S. cerevisiae resulted in the

production of small amounts of 1,2-PDO (Hoffman, 1999; Jeon et al., 2009; Lee and DaSilva, 2006).

Nevertheless, the additional expression of E. coli gldA encoding a glycerol dehydrogenase was

demonstrated to increase 1,2-PDO production in S. cerevisiae (Hoffman, 1999). The E. coli aldehyde

reductase encoded by yqhD was also considered here to be a useful enzyme for converting

methylglyoxal to 1,2-PDO. The enzyme was shown to reduce methylglyoxal in vitro (Jarboe, 2011; Lee

21

et al., 2010) and has been successfully used to improve 1,2-PDO formation in engineered

C. glutamicum additionally expressing E. coli mgsA and gldA (Siebert and Wendisch, 2015).

Therefore, we decided to use codon-optimized versions of the E. coli genes mgsA, yqhD and gldA

(referred to as EcmgsA, EcyqhD and EcgldA, respectively) in the current study and placed them

between different S. cerevisiae promoters and terminators (see Material and Methods) in order to

establish a `1,2-PDO module´ in yeast. All engineered strains harboring this module (referred to as

`PDO´) carried head-to-tail genomic integrations of S. cerevisiae expression cassettes for EcmgsA and

EcgldA, while the expression cassette for EcyqhD was expressed from a S. cerevisiae low-copy

expression vector (see section 2.5 and Table 1).

While establishing the module `PDO´, a parallel study in our laboratory revealed that the expression

of a heterologous glycerol facilitator significantly improves the growth of strain CBS 6412-13A in

synthetic glycerol medium (Klein et al., 2016b). As we assumed that an improved glycerol uptake

might also be beneficial for 1,2-PDO production from glycerol, we decided to integrate an expression

cassette bearing the coding sequence of FPS1 from Cyberlindnera jadinii (referred to here as CjFPS1)

into the genome of all strains engineered for 1,2-PDO production. Accordingly, the term `FPS´ was

included in the respective strain names (see section 2.5 and Table 1). The strain CBS 6412-13A

expressing CjFPS1 was used as the negative reference throughout this study but CjFPS1 expression

was not considered here as a separate module.

A second module addressed the co-factor supply for 1,2-PDO formation. The three step conversion

from DHAP to 1,2-PDO includes two enzymatic reactions that require reducing equivalents in the

form of cytosolic NAD(P)H (Figure 1). As mentioned in the introduction (section 1), the native FAD-

dependent glycerol catabolic pathway (up to DHAP) in S. cerevisiae does not deliver cytosolic

reducing equivalents. However, the pathway can be functionally replaced by an NAD+-dependent

`DHA pathway´ as recently demonstrated by Klein et al. (2016a). The respective genetic modifications

included the plasmid-based heterologous expression of glycerol dehydrogenase from Ogataea

parapolymorpha (Opgdh), the overexpression of endogenous dihydroxyacetone kinase (DAK1) and

22

the deletion of endogenous glycerol kinase (GUT1). These modifications are collectively referred to

as module `DHA´ in the respective strain names (Table 1). This module has been supposed to strongly

increase cytosolic NADH availability for 1,2-PDO production since the FAD-dependent native pathway

for glycerol catabolism has been completely abolished.

An additional module addressed the need for increased availability of DHAP, the metabolic precursor

for the `methylglyoxal pathway´. The introduction of the heterologous 1,2-PDO pathway creates a

metabolic branch point at which MGS must compete with the endogenous glycolytic enzyme

triosephosphate isomerase (TPI) (encoded by TPI1) for their common substrate DHAP. Jung et al.

(2008) showed that the deletion of the TPI1 gene in S. cerevisiae bearing a heterologous 1,2-PDO

pathway resulted in increased 1,2-PDO production in a medium containing a mix of glucose,

galactose and ethanol as carbon sources. However, the complete abolishment of TPI activity is not

feasible when glycerol serves as the sole source of carbon since this enzyme is essential to channel

the carbon flux from the substrate glycerol into the central carbon metabolism (i.e. both glycolysis

and gluconeogenesis). Therefore, a strategy for downregulating TPI1 expression (without fully

abolishing it) via promoter replacement was employed. In a previous study, a collection of mutant

promoters with gradually decreasing strengths (based on the constitutive TEF1 promoter) had been

generated (Alper et al., 2005). Nevoigt et al. (2006) used this collection to develop sets of promoter

replacement cassettes allowing the exchange of any promoter in the S. cerevisiae genome with a

TEF1 promoter variant of defined strength. Two weak promoters from this collection (TEFmut7 and

TEFmut2) have initially been tested in the current study (Figure S2 and the respective supplementary

information), and we eventually selected the weakest promoter (TEFmut2) in order to replace the

native TPI1 promoter. The respective genetic modification is referred to here as module `TPI1down´

(Table 1). The impact of the promoter replacement on growth of the corresponding intermediate

strain in synthetic medium containing different carbon sources is shown in Figure S2 (supplementary

information). These results are an indirect proof for the fact that the promoter replacement (using

the TEFmut2 promoter) indeed resulted in decreased TPI activity.

23

3.2 The choice of cultivation medium strongly affected 1,2-PDO production of the engineered

strains

Our initial intention was to use synthetic glycerol medium for studying the impacts of the different

engineering modules on 1,2-PDO production. We decided to use urea as the source of nitrogen in our

synthetic medium as it, in contrast to ammonium sulfate, does not lead to medium acidification and

is therefore known to result in much higher final optical densities compared to synthetic medium

with ammonium sulfate. In the urea-containing medium, the strain PDO-FPS-TPI1down-DHA produced

less than 0.2 g L-1 after 120 h of cultivation in synthetic glycerol medium (Figure 2A). Instead,

significant concentrations of ethanol indicating an onset of fermentative metabolism were detected

in the first half of the cultivation with a peak of up to 18 g L-1 after 72 h. Later, the ethanol

concentration decreased (Figure 2B).

In a previous study, Jung et al. (2011) successfully demonstrated the production of 1,2-PDO in

engineered S. cerevisiae when using a complex medium (YPG containing 10 g L-1 yeast extract, 20 g L-1

peptone, 10 g L-1glycerol and 1 g L-1 galactose). We therefore tested our advanced strain PDO-FPS-

TPI1down-DHA in YPG6% medium (10 g L-1 yeast extract, 20 g L-1 peptone and 60 mL L-1 glycerol).

However, only a slight improvement of 1,2-PDO production could be observed compared to synthetic

medium with urea. The highest detected titer was ~0.3 g L-1 after 96 h of cultivation. Notably, the

switch to YPG6% medium resulted in faster growth, glycerol consumption as well as ethanol

production (Figure 2). Under these conditions, the strain accumulated up to 18 g L-1 ethanol already

during the first 24 h of cultivation (Figure 2B).

Surprisingly, omitting peptone from our YPG6% medium remarkably increased 1,2-PDO production

from ~0.3 g L-1 to ~1.3 g L-1 where most of the product accumulated during the first 48 h (Figure 2A).

The plausible nutrient limitation caused a much slower growth and is also reflected by a slower rate

of glycerol consumption. Surprisingly, much less ethanol was produced in the absence of peptone

(Figure 2B). On the basis of these results, we decided to first characterize the influence of the

24

different metabolic engineering modules on 1,2-PDO production in YG6% medium and dedicated

separate experiments scrutinizing the medium effects on 1,2-PDO production.

Figure 2 (2 column fitting image). Physiological characterization of strain PDO-FPS-TPI1down-DHA in three

different cultivation media containing 60 mL L-1 glycerol as the sole carbon source. The strain carried all

engineering modules as described in section 3.1. Cultivations were performed in 50 mL shake flask cultures and

culture supernatants were analyzed by HPLC for the production of 1,2-propanediol (1,2-PDO) (A), ethanol (B)

and the consumption of glycerol (C). Growth was recorded by determining the optical density of the cultures at

a wavelength of 600 nm (OD600) (D). The different media used were SMG (synthetic defined glycerol medium

containing urea as the nitrogen source), YPG (complex glycerol medium containing 1% yeast extract and 2%

peptone) and YG (complex glycerol medium containing 1% yeast extract). All mean values and standard

deviations are derived from at least three biological replicates.

25

3.3 The combination of modules DHA and TPI1down was crucial to achieve the highest concentration

of 1,2-PDO in YG6% medium

As mentioned in section 3.2, YG6% medium was used in order to characterize all constructed strains

and dissect the impacts of the individual modules and their combinations on 1,2-PDO production and

the cells’ physiology (Figure 3).

Figure 3 (2 column fitting image). Physiological characterization of S. cerevisiae strains (derivatives of

CBS 6412-13A) engineered for 1,2-PDO production from glycerol in YG medium containing 60 mL L-1 glycerol.

Besides an expression cassette for the glycerol facilitator encoded by CjFPS1 (`FPS´), the analyzed strains

carried the engineering module `PDO´ alone or in combination with the modules `DHA´ and `TPI1down´ as

described in section 3.1. Cultivations were performed in 50 mL shake flask cultures and culture supernatants

were analyzed by HPLC for the production of 1,2-PDO (A), ethanol (B) and the consumption of glycerol (C).

26

Growth was recorded by optical density measurements at 600 nm (D). Mean values and standard deviations

were determined from at least three biological replicates.

Compared to the control strain FPS, all strains bearing the `1,2-PDO module´ produced at least small

amounts of 1,2-PDO (Figure 3 A). Interestingly, neither the sole downregulation of TPI1 expression

(module `TPI1down´) nor the sole glycerol catabolic pathway replacement (module `DHA´) in the

genetic background of the strain PDO-FPS resulted in a significant improvement of the 1,2-PDO titer.

Only the combination of these two metabolic engineering modules with the `1,2-PDO module´

considerably improved 1,2-PDO production from less than 0.2 g L-1 to ~1.3 g L-1 when measured after

120 h of cultivation. The results demonstrate that both above-mentioned modules were essential for

1,2-PDO production from glycerol.

The glycerol catabolic pathway replacement in strain PDO-FPS improved the strain’s glycerol

consumption (Figure 3C, strains PDO-FPS-DHA and PDO-FPS-TPI1down-DHA) indicating that the DHA

pathway allows a better flux from glycerol to DHAP compared to the G3P pathway present in the

wild-type strain. In strain PDO-FPS-DHA, the increased glycerol consumption coincides with a faster

growth, while the presence of the module `TPI1down´ seems to abrogate the growth supporting effect

of module `DHA´ as visible in the growth curve for strain PDO-FPS-TPI1down-DHA (Figure 3D).

Our data clearly show that the genetic manipulation mainly responsible for the production of ethanol

observed in YG6% medium is the replacement of the endogenous glycerol catabolic pathway by the

NADH-generating DHA pathway (Figure 3B). A switch to fermentative metabolism (visible by ethanol

formation) is plausible when considering the excess of cytosolic NADH generated by the latter

pathway. Again, the ethanol production was particularly high in the first half of the cultivation

period, while its concentration declined during the second part of the cultivation (Figure 3B). During

the second phase of cultivation, glycerol was still present and its concentration continued to

decrease. Interestingly, the superior growth of strain PDO-FPS-DHA coincided with a strong decrease

in ethanol concentration while the respective reduction in ethanol concentration during the second

27

cultivation half was very low in the strain with additional downregulation of TPI1 expression

(Figure 3B).

3.4 Implementation of a second expression cassette for E. coli MGS and analysis in controlled batch

cultivation in bioreactors

In the previous section, it was shown that the modules for cofactor-supply and carbon flux re-

direction (`DHA´ and `TPI1down´, respectively) were essential genetic manipulations to increase the

capacity of S. cerevisiae to produce 1,2-PDO from glycerol. The next question that arose was whether

the flux into the 1,2-PDO pathway was rate-controlling for 1,2-PDO production in the strain PDO-FPS-

TPI1down-DHA. At this stage, we decided to construct a strain that carries all genetic modifications

within its genome (including the implementation of an additional expression cassette for E. coli

mgsA) resulting in strain PDO-FPS-TPI1down-DHA EcmgsA2. This also facilitated the physiological

characterization in larger volumes since no antibiotic had to be added anymore. When this strain was

analyzed in shake flasks under the same conditions as used before, the additional EcmgsA expression

cassette resulted in increased 1,2-PDO production (~2.5 g L-1 after 140 h of cultivation) while the

maximum ethanol titer decreased from ~6.6 g L-1 to ~3.8 g L-1 as shown in Figure S3 (supplementary

information).

As a next step, the strain PDO-FPS-TPI1down-DHA EcmgsA2 was characterized under controlled

conditions in batch cultivations in bioreactors using the same medium (YG6%). In general, the strain’s

physiology was very similar to that observed in shake flask cultures. However, the controlled

cultivation conditions allowed the cells to reach higher optical densities than in shake flasks. The

maximum 1,2-PDO titer reached was ~4.3 g L-1 (Figure 4). This has been the highest reported 1,2-PDO

titer obtained by fermentations with engineered S. cerevisiae.

28

Figure 4 (single column fitting image). Physiological characterization of strain PDO-FPS-TPI1down-DHA-

EcmgsA2 in batch cultivations in 1 L bioreactors using YG medium containing 60 mL L-1 glycerol as the carbon

source. The analyzed strain carried genomic integrations of all engineering modules, i.e. besides the glycerol

facilitator encoded by CjFPS1 (`FPS´), the methylglyoxal pathway (`PDO´) including a second expression cassette

for E. coli mgsA (EcmgsA2), the glycerol catabolic pathway replacement (`DHA´) and the replacement of the

native TPI1 promoter by the weak TEF1 promoter variant TEFmut2 (`TPI1down´). One representative out of three

experiments is shown.

3.5 1,2-PDO production in buffered synthetic medium with ammonium sulfate as a carbon source

The rather surprising results concerning the influence of the medium composition on the observable

product pattern encouraged us to conduct a number of experiments in order to gain further insight

into this phenomenon. The results (data not shown) revealed that the impact of medium

components is very complex and most probably even strain dependent. Further experiments have to

be carried out in order to dissect and understand the observed medium effects. Among the already

performed experiments regarding medium effects, there has been one that was set to test whether

glycerol fermentation is also visible in synthetic glycerol medium with ammonium sulfate as the

nitrogen source as soon as a buffer is added and thereby rapid medium acidification is avoided

resulting in significantly higher optical densities. This question came up after we detected significant

ethanol production in our strains with the `DHA´ module (sections 3.2 and 3.3), a fact that has not

29

been recognized in our previous study regarding DHA pathway strains (Klein et al., 2016a). However,

the earlier study was performed using unbuffered synthetic medium with ammonium sulfate as the

nitrogen source. Therefore, 100 mM potassium phthalate buffer was added to synthetic medium

containing either ammonium sulfate or urea. This allowed a direct comparison of the two nitrogen

sources. The respective experiment was performed with the above-mentioned strain PDO-FPS-

TPI1down-DHA EcmgsA2 carrying two expression cassettes for EcmgsA and in which all genetic

modifications were stably integrated into the genome. To our huge surprise, the buffered

ammonium sulfate containing medium led to a significant formation of 1,2-PDO and a comparably

low ethanol production which was in clear contrast to the results obtained for the urea-containing

medium (Figure 5). The remarkable difference with regard to the 1,2-PDO/ethanol ratio between the

two nitrogen sources (Figure 5A and B) cannot simply be explained by the obtained culture density

(Figure 5A). In fact, the used nitrogen source seemed to have a (so far unknown) impact on the

carbon flux distribution in our engineered strains.

30

Figure 5 (2 column fitting image). Influence of the used nitrogen source on the physiology of the S. cerevisiae

strain PDO-FPS-TPI1down-DHA-EcmgsA2 engineered for 1,2-PDO production from glycerol in synthetic medium

buffered with 100 mM potassium phthalate buffer. The initial pH of the media was set to 5.0. Cultivations

were performed in 50 mL shake flask cultures and culture supernatants were analyzed by HPLC for the

production of 1,2-PDO (A), ethanol (B) and the consumption of glycerol (C). Growth was recorded by optical

density measurements at 600 nm (D). Mean values and standard deviations were determined from at least

three biological replicates.

31

4. Discussion

In this study, a novel metabolic engineering approach was used to produce 1,2-PDO in S. cerevisiae

from glycerol as the sole carbon source. A previous attempt to establish 1,2-PDO production from

glycerol in S. cerevisiae (equipped with the ‘methylglyoxal pathway’) focused on the overexpression

of GUT1 and GUT2 which encode the enzymes for the native (FAD-dependent) glycerol catabolic

pathway (Jung et al., 2011). However, the electrons stemming from glycerol oxidation through this

pathway are eventually transferred to oxygen via the respiratory chain. In contrast, our current

approach was based on the complete replacement of the endogenous G3P pathway by the NAD+-

dependent DHA pathway according to Klein et al. (2016a). This replacement has been supposed to

provide cytosolic reducing equivalents from glycerol and facilitate the formation of fermentation

products such as 1,2-PDO. The current study demonstrates that the module `DHA´ indeed increased

the formation of the target product at least when TPI expression was reduced in parallel thereby

increasing the availability of the precursor DHAP. Notably, downregulation of TPI expression alone

did not result in significant 1,2-PDO formation. These results underline the importance of providing

sufficient concentrations of both pre-cursor and co-factor for the successful re-direction of metabolic

fluxes. The fact that 1,2-PDO production in the presence of increased pre-cursor and co-factor could

be even further increased by a second mgsA expression cassette demonstrates that the flux through

methylglyoxal synthase has been the rate-controlling step in the respective genetic background and

might be one important target for further strain improvement.

As already mentioned, the 1,2-PDO titer of ˃ 4 g/l achieved in the current study has been the highest

reported for S. cerevisiae. This titer obtained in the bioreactor experiment in YG medium

corresponded to a maximum product yield of ~129 ± 21 mg per g glycerol consumed. Jung et al.

(2011) reported a 1,2-PDO yield of 213 mg per g glycerol consumed. However, their experiments

were performed in modified YEPD medium (yeast extract 1%, peptone 2%, glycerol 1% and galactose

0.1%). Although the galactose was only added for induction of gene expression in the latter study, its

utilization as a carbon source may have falsified the calculated yields. Moreover, ingredients of the

32

added peptone might have also been used as minor carbon sources. These facts hamper a reliable

comparison with our results.

Regarding the pathway from DHAP to 1,2-PDO, there are several possible metabolic routes which

may result in either R- or S-1,2-PDO (Figure 1). However, we have not dissected in our analytics so far

whether only one or both enantiomers were formed by our engineered S. cerevisiae strains. In

addition, it has been difficult to draw a strong hypothesis on the actual carbon flux solely based on

published knowledge. First, the genes EcgldA and EcyqhD encode for enzymes with relatively broad

substrate specificities (Jarboe, 2011; St. Martin et al., 1977; Tang et al., 1979). Second, the gene

Opgdh originally used for the module ‘DHA’ might also contribute to the reductive reactions from

methylglyoxal to 1,2-PDO to a certain degree since its expression (instead of EcgldA) in a PDOmin

strain also increased 1,2-PDO formation as compared to the sole expression of EcmgsA (results not

shown). Third, endogenous (probably rather unspecific) reductases of the yeast S. cerevisiae have

also to be considered to be active in the respective chemical conversions as mentioned in section 1.

Nevertheless, our 1,2-PDO producing strains formed a significant amount of an additional product as

visible by a conspicuous peak in the HPLC (data not shown). HPLC analysis of the retention times of

the pathway intermediates methylglyoxal, R-/S-lactaldehyde and acetol indicated that acetol might

be the compound that was accumulated by our 1,2-PDO producer. The retention time of

lactaldehyde under the employed conditions is close to that of glycerol and the latter was still

present in relatively large concentrations in the medium at the end of the cultivation period in most

of our experiments. Therefore, a peak for lactaldehyde is assumed to partially overlap with the peak

for glycerol. At least, we did not recognize any shoulder for the glycerol peak in our chromatograms.

Notably, other authors expressing the same E. coli genes for 1,2-PDO formation in C. glutamicum also

reported significant acetol formation (Siebert and Wendisch, 2015), while Altaras and Cameron

(1999) found R-lactaldehyde as an intermediate in an E. coli strain overexpressing E. coli mgsA and

gldA. Future work including metabolomics will focus on a more comprehensive characterization of

the carbon flow in our engineered strains.

33

Glycerol has been considered as a carbon source that is solely metabolized in a respiratory manner in

S. cerevisiae. The full dependence of glycerol catabolism on respiration in wild-type S. cerevisiae

strains can be explained by the fact that the electrons are directly transferred by Gut2 via FADH2 to

the mitochondrial respiratory chain. This also holds for strains which display higher growth rates (and

glycolytic flux) in synthetic glycerol medium than the majority of wild-type S. cerevisiae strains. In

fact, Ochoa-Estopier et al. (2011) comprehensively studied an evolved derivative of CBS 8066 (FL20)

that exhibited a µmax of 0.20 h-1 in synthetic glycerol medium in a controlled bioreactor setup and no

reduced fermentation products such as ethanol were detected. Similarly, no fermentation products

were detected in shake-flask cultivations using our non-engineered wild-type S. cerevisiae strain

CBS 6412-13A which is able to grow in synthetic glycerol medium with a µmax of 0.13 h-1 via the G3P

pathway (data not shown).

In the view of the current study it is, however, important to note that the applied glycerol catabolic

pathway replacement strategy rendered the strain’s glycerol catabolism solely dependent on NAD+.

This has been assumed to result in surplus cytosolic NADH and to facilitate the production of

fermentation products in our strains such as 1,2-PDO (heterologous pathway) and/or ethanol (S.

cerevisiae’s major native fermentation product). Indeed, our data indicates that cells, which depend

on the DHA pathway for glycerol catabolism, at least partially ferment the glycerol, even though both

the degree of fermentation and the ratio between 1,2-PDO and ethanol formation are strain- and

medium dependent.

So far, we cannot dissect whether the surplus of cytosolic NADH caused the fermentative

metabolism of glycerol right at the beginning of the cultivation or whether any kind of nutrient or

oxygen limitation was an additional requirement for the metabolic switch. At least, one has to note

that ethanol formation was also visible in the bioreactor experiment employing the YG6% medium

where air was flushed with a rate of 1 vvm (stirring rate 800 rpm) into the culture. The underlying

basis why 1,2-PDO is produced in significant amounts in YG6% medium (and in the buffered

ammonium sulfate-containing synthetic medium) while only trace amounts of the target product

34

were formed in synthetic medium with urea as a nitrogen source and in YPG6% medium (where

ethanol is the prominent fermentation product) has not been clarified yet. By unknown reasons, the

engineered 1,2-PDO pathway can probably better compete with the endogenous ethanol formation

pathway in the first two media.

It has been very intriguing that the concentration of ethanol in the medium started to decrease at a

certain point in time. In general, the actual ethanol concentration could have been affected by three

different processes which might, at least partly, have occurred in parallel: i) ethanol formation, ii)

evaporation and iii) ethanol consumption. Evaporation was tested in a parallel experiment using

medium without cells and an initial ethanol concentration of 25 g L-1. Under the applied conditions,

the ethanol evaporated with a maximum rate of ~0.10 g L-1 h-1. This rate was lower than observed for

the decrease in ethanol concentration for the analyzed cultures which varied between 0.16 and 0.44

g L-1 h-1. Although its impact can definitely not be neglected, evaporation can therefore not

completely explain the observed decrease in ethanol concentration in our experiments. Thus, it can

be concluded that ethanol might have been consumed together with glycerol and even further when

glycerol was completely depleted (Figure 2). Assuming the actual occurrence of ethanol consumption

by the cells, the observed decrease of the ethanol titer in the second cultivation period may have

been caused by the fact that ethanol production by fermentation stopped (or at least slowed down)

while respiratory ethanol utilization continued or even increased. A reduction in glycerol

consumption rate at a later stage in the cultivation was also visible. Although we do not have a

plausible explanation for this result, it might underpin our hypothesis with regard to a certain

overflow metabolism at increased glycerol consumption rates. The lower glycerol consumption rate

might have resulted in less NADH generation and the rate of respiration might have been sufficient

to reoxidize the formed NADH explaining why the cells did not need fermentation at the later phase

of cultivation. Nevertheless, we cannot exclude that regulatory mechanisms changed the rate of

fermentation and/or respiration at this point in time and further work will be necessary to clarify this

aspect.

35

The data presented in the current manuscript demonstrate that S. cerevisiae can be engineered to

produce significant concentrations of 1,2-PDO (> 4 g L-1 in a controlled bioreactor set-up). Our study

shows that both metabolic precursor and co-factor supply are essential for target product formation.

The work also demonstrates the crucial impact of the medium composition and the experimental

conditions on the metabolic mode and, thus product titer and yield. We therefore believe that there

is still a lot of room for further improving 1,2-PDO production in S. cerevisiae.

5. Acknowledgements

This work was funded through the ERA-NET scheme of the 7th EU Framework Program (IPCRES;

German Federal Ministry of Education and Research, Project No. 031A344). We thank Johan M.

Thevelein (KU Leuven, Belgium) for kindly providing us with the plasmid pGG119 and Solvejg

Sevecke, Martina Carrillo, and Joeline Xiberras for technical support.

36

REFERENCES

Akada, R., Hirosawa, I., Kawahata, M., Hoshida, H., Nishizawa, Y., 2002. Sets of integrating plasmids and gene disruption cassettes containing improved counter-selection markers designed for repeated use in budding yeast. Yeast. 19, 393-402.

Alper, H., Fischer, C., Nevoigt, E., Stephanopoulos, G., 2005. Tuning genetic control through promoter engineering. Proc Natl Acad Sci U S A. 102, 12678-83.

Altaras, N. E., Cameron, D. C., 1999. Metabolic engineering of a 1,2-propanediol pathway in Escherichia coli. Appl Environ Microbiol. 65, 1180-5.

Altaras, N. E., Cameron, D. C., 2000. Enhanced production of (R)-1,2-propanediol by metabolically engineered Escherichia coli. Biotechnol Prog. 16, 940-6.

Badia, J., Ros, J., Aguilar, J., 1985. Fermentation mechanism of fucose and rhamnose in Salmonella typhimurium and Klebsiella pneumoniae. J Bacteriol. 161, 435-7.

Barbier, G. G., Ladd, J. L., Campbell, E. R., 2011. Genetic Modification of Pichia pastoris for Production of Propylene Glycol from Glycerol. International Journal of Genetic Engineering. 1, 6-13.

Bennett, G. N., San, K. Y., 2001. Microbial formation, biotechnological production and applications of 1,2-propanediol. Appl Microbiol Biotechnol. 55, 1-9.

Berrios-Rivera, S. J., San, K. Y., Bennett, G. N., 2003. The effect of carbon sources and lactate dehydrogenase deletion on 1,2-propanediol production in Escherichia coli. J Ind Microbiol Biotechnol. 30, 34-40.

Borodina, I., Nielsen, J., 2014. Advances in metabolic engineering of yeast Saccharomyces cerevisiae for production of chemicals. Biotechnol J. 9, 609-20.

Boronat, A., Aguilar, J., 1979. Rhamnose-induced propanediol oxidoreductase in Escherichia coli: purification, properties, and comparison with the fucose-induced enzyme. J Bacteriol. 140, 320-6.

Boronat, A., Aguilar, J., 1981. Metabolism of L-fucose and L-rhamnose in Escherichia coli: differences in induction of propanediol oxidoreductase. J Bacteriol. 147, 181-5.

Burk, M. J., Van Dien, S., 2016. Biotechnology for Chemical Production: Challenges and Opportunities. Trends Biotechnol. 34, 187-90.

Cameron, D. C., Altaras, N. E., Hoffman, M. L., Shaw, A. J., 1998. Metabolic engineering of propanediol pathways. Biotechnol Prog. 14, 116-25.

Cameron, D. C., Cooney, C. L., 1986. A Novel Fermentation - the Production of R(-)-1,2-Propanediol and Acetol by Clostridium-Thermosaccharolyticum. Bio-Technology. 4, 651-654.

Chao, R., Yuan, Y., Zhao, H., 2015. Recent advances in DNA assembly technologies. FEMS Yeast Res. 15, 1-9.

Clomburg, J. M., Gonzalez, R., 2011. Metabolic engineering of Escherichia coli for the production of 1,2-propanediol from glycerol. Biotechnol Bioeng. 108, 867-79.

da Silva, G. P., Mack, M., Contiero, J., 2009. Glycerol: a promising and abundant carbon source for industrial microbiology. Biotechnol Adv. 27, 30-9.

David, F., Siewers, V., 2015. Advances in yeast genome engineering. FEMS Yeast Res. 15, 1-14. Dowd, M. K., Johansen, S. L., Cantarella, L., Reilly, P. J., 1994. Low-Molecular-Weight Organic

Composition of Ethanol Stillage from Sugarcane Molasses, Citrus Waste, and Sweet Whey. Journal of Agricultural and Food Chemistry. 42, 283-288.

Flagfeldt, D. B., Siewers, V., Huang, L., Nielsen, J., 2009. Characterization of chromosomal integration sites for heterologous gene expression in Saccharomyces cerevisiae. Yeast. 26, 545-51.

Gibson, D. G., Young, L., Chuang, R. Y., Venter, J. C., Hutchison, C. A., 3rd, Smith, H. O., 2009. Enzymatic assembly of DNA molecules up to several hundred kilobases. Nat Methods. 6, 343-5.

Gietz, R. D., Schiestl, R. H., Willems, A. R., Woods, R. A., 1995. Studies on the transformation of intact yeast cells by the LiAc/SS-DNA/PEG procedure. Yeast. 11, 355-60.

37

Gonzalez, R., Murarka, A., Dharmadi, Y., Yazdani, S. S., 2008. A new model for the anaerobic fermentation of glycerol in enteric bacteria: trunk and auxiliary pathways in Escherichia coli. Metab Eng. 10, 234-45.

Gueldener, U., Heck, S., Fielder, T., Beinhauer, J., Hegemann, J. H., 1996. A new efficient gene disruption cassette for repeated use in budding yeast. Nucleic Acids Res. 24, 2519-24.

Gueldener, U., Heinisch, J., Koehler, G. J., Voss, D., Hegemann, J. H., 2002. A second set of loxP marker cassettes for Cre-mediated multiple gene knockouts in budding yeast. Nucleic Acids Res. 30, e23.

Hoffman, C. S., Winston, F., 1987. A ten-minute DNA preparation from yeast efficiently releases autonomous plasmids for transformation of Escherichia coli. Gene. 57, 267-72.

Hoffman, M. L., Metabolic engineering of 1,2-Propanediol production in Saccharomyces cerevisiae. University of Wisconsin--Madison, 1999, pp. xi, 218 leaves.

Hong, K. K., Nielsen, J., 2012. Metabolic engineering of Saccharomyces cerevisiae: a key cell factory platform for future biorefineries. Cell Mol Life Sci. 69, 2671-90.

Huang, K., Rudolph, F. B., Bennett, G. N., 1999. Characterization of methylglyoxal synthase from Clostridium acetobutylicum ATCC 824 and its use in the formation of 1, 2-propanediol. Appl Environ Microbiol. 65, 3244-7.

Jain, R., Huang, J., Yuan, Q., Yan, Y., 2015a. Engineering microaerobic metabolism of E. coli for 1,2-propanediol production. J Ind Microbiol Biotechnol. 42, 1049-55.

Jain, R., Sun, X., Yuan, Q., Yan, Y., 2015b. Systematically engineering Escherichia coli for enhanced production of 1,2-propanediol and 1-propanol. ACS Synth Biol. 4, 746-56.

Jarboe, L. R., 2011. YqhD: a broad-substrate range aldehyde reductase with various applications in production of biorenewable fuels and chemicals. Appl Microbiol Biotechnol. 89, 249-57.

Jeon, E., Lee, S., Kim, D., Yoon, H., Oh, M., Park, C., Lee, J., 2009. Development of a Saccharomyces cerevisiae strain for the production of 1,2-propanediol by gene manipulation. Enzyme and Microbial Technology. 45, 42-47.

Johnson, D. T., Taconi, K. A., 2007. The glycerin glut: Options for the value-added conversion of crude glycerol resulting from biodiesel production. Environmental Progress. 26, 338-348.

Jung, J. Y., Choi, E. S., Oh, M. K., 2008. Enhanced production of 1,2-propanediol by tpi1 deletion in Saccharomyces cerevisiae. J Microbiol Biotechnol. 18, 1797-802.

Jung, J. Y., Yun, H. S., Lee, J., Oh, M. K., 2011. Production of 1,2-propanediol from glycerol in Saccharomyces cerevisiae. J Microbiol Biotechnol. 21, 846-53.

Karinen, R. S., Krause, A. O. I., 2006. New biocomponents from glycerol. Applied Catalysis A: General. 306, 128-133.

Klein, M., Carrillo, M., Xiberras, J., Islam, Z. U., Swinnen, S., Nevoigt, E., 2016a. Towards the exploitation of glycerol's high reducing power in Saccharomyces cerevisiae-based bioprocesses. Metab Eng.

Klein, M., Islam, Z.-u., Knudsen, P. B., Carrillo, M., Swinnen, S., Workman, M., Nevoigt, E., 2016b. The expression of glycerol facilitators from various yeast species improves growth on glycerol of Saccharomyces cerevisiae. Metabolic Engineering Communications. 3, 252-257.

Klein, M., Swinnen, S., Thevelein, J., Nevoigt, E., 2016c. Glycerol metabolism and transport in yeast and fungi: Established knowledge and ambiguities. Environ Microbiol.

Lee, C., Kim, I., Lee, J., Lee, K. L., Min, B., Park, C., 2010. Transcriptional activation of the aldehyde reductase YqhD by YqhC and its implication in glyoxal metabolism of Escherichia coli K-12. J Bacteriol. 192, 4205-14.

Lee, M. J., Brown, I. R., Juodeikis, R., Frank, S., Warren, M. J., 2016. Employing bacterial microcompartment technology to engineer a shell-free enzyme-aggregate for enhanced 1,2-propanediol production in Escherichia coli. Metab Eng. 36, 48-56.

Lee, W., DaSilva, N. A., 2006. Application of sequential integration for metabolic engineering of 1,2-propanediol production in yeast. Metab Eng. 8, 58-65.

Li, H., Liao, J. C., 2013. Engineering a cyanobacterium as the catalyst for the photosynthetic conversion of CO2 to 1,2-propanediol. Microb Cell Fact. 12, 4.

38

Mattam, A. J., Clomburg, J. M., Gonzalez, R., Yazdani, S. S., 2013. Fermentation of glycerol and production of valuable chemical and biofuel molecules. Biotechnol Lett. 35, 831-42.

Merico, A., Ragni, E., Galafassi, S., Popolo, L., Compagno, C., 2011. Generation of an evolved Saccharomyces cerevisiae strain with a high freeze tolerance and an improved ability to grow on glycerol. J Ind Microbiol Biotechnol. 38, 1037-44.

Murarka, A., Dharmadi, Y., Yazdani, S. S., Gonzalez, R., 2008. Fermentative utilization of glycerol by Escherichia coli and its implications for the production of fuels and chemicals. Appl Environ Microbiol. 74, 1124-35.

Naito, Y., Hino, K., Bono, H., Ui-Tei, K., 2015. CRISPRdirect: software for designing CRISPR/Cas guide RNA with reduced off-target sites. Bioinformatics. 31, 1120-3.

Nevoigt, E., 2008. Progress in metabolic engineering of Saccharomyces cerevisiae. Microbiol Mol Biol Rev. 72, 379-412.

Nevoigt, E., Genetically modified yeast with improved glycerol catabolism. Vol. DE112015003261 (A5), Germany, 2017.

Nevoigt, E., Kohnke, J., Fischer, C. R., Alper, H., Stahl, U., Stephanopoulos, G., 2006. Engineering of promoter replacement cassettes for fine-tuning of gene expression in Saccharomyces cerevisiae. Appl Environ Microbiol. 72, 5266-73.

Niimi, S., Suzuki, N., Inui, M., Yukawa, H., 2011. Metabolic engineering of 1,2-propanediol pathways in Corynebacterium glutamicum. Appl Microbiol Biotechnol. 90, 1721-9.

Ochoa-Estopier, A., Lesage, J., Gorret, N., Guillouet, S. E., 2011. Kinetic analysis of a Saccharomyces cerevisiae strain adapted for improved growth on glycerol: Implications for the development of yeast bioprocesses on glycerol. Bioresour Technol. 102, 1521-7.

Shelley, S., 2007. A renewable route to propylene glycol. Chem Eng Prog. 103, 6-9. Siebert, D., Wendisch, V. F., 2015. Metabolic pathway engineering for production of 1,2-propanediol

and 1-propanol by Corynebacterium glutamicum. Biotechnol Biofuels. 8, 91. St. Martin, E. J., Freedberg, W. B., Lin, E. C., 1977. Kinase replacement by a dehydrogenase for

Escherichia coli glycerol utilization. J Bacteriol. 131, 1026-8. Steensma, H. Y., Ter Linde, J. J., 2001. Plasmids with the Cre-recombinase and the dominant nat

marker, suitable for use in prototrophic strains of Saccharomyces cerevisiae and Kluyveromyces lactis. Yeast. 18, 469-72.

Suzuki, T., Onishi, H., 1968. Aerobic Dissimilation of L-Rhamnose and Production of L-Rhamnonic Acid and 1,2-Propanediol by Yeasts. Agricultural and Biological Chemistry. 32, 888-&.

Swinnen, S., Ho, P. W., Klein, M., Nevoigt, E., 2016. Genetic determinants for enhanced glycerol growth of Saccharomyces cerevisiae. Metab Eng. 36, 68-79.

Swinnen, S., Klein, M., Carrillo, M., McInnes, J., Nguyen, H. T., Nevoigt, E., 2013. Re-evaluation of glycerol utilization in Saccharomyces cerevisiae: characterization of an isolate that grows on glycerol without supporting supplements. Biotechnol Biofuels. 6, 157.

Tang, C. T., Ruch, F. E., Jr., Lin, C. C., 1979. Purification and properties of a nicotinamide adenine dinucleotide-linked dehydrogenase that serves an Escherichia coli mutant for glycerol catabolism. J Bacteriol. 140, 182-7.

Thompson, J. C., He, B. B., 2006. Characterization of crude glycerol from biodiesel production from multiple feedstocks. Applied Engineering in Agriculture. 22, 261-265.

Tran-Din, K., Gottschalk, G., 1985. Formation of d(-)-1,2-propanediol and d(-)-lactate from glucose by Clostridium sphenoides under phosphate limitation. Archives of Microbiology. 142, 87-92.

Turner, K. W., Roberton, A. M., 1979. Xylose, arabinose, and rhamnose fermentation by Bacteroides ruminicola. Appl Environ Microbiol. 38, 7-12.

Verduyn, C., Postma, E., Scheffers, W. A., Van Dijken, J. P., 1992. Effect of benzoic acid on metabolic fluxes in yeasts: a continuous-culture study on the regulation of respiration and alcoholic fermentation. Yeast. 8, 501-17.

Yang, F., Hanna, M. A., Sun, R., 2012. Value-added uses for crude glycerol--a byproduct of biodiesel production. Biotechnol Biofuels. 5, 13.

39

Yazdani, S. S., Gonzalez, R., 2007. Anaerobic fermentation of glycerol: a path to economic viability for the biofuels industry. Curr Opin Biotechnol. 18, 213-9.

Yazdani, S. S., Gonzalez, R., 2008. Engineering Escherichia coli for the efficient conversion of glycerol to ethanol and co-products. Metabolic Engineering. 10, 340-351.

Zhang, L., Tang, Y., Guo, Z., Shi, G., 2013. Engineering of the glycerol decomposition pathway and cofactor regulation in an industrial yeast improves ethanol production. J Ind Microbiol Biotechnol. 40, 1153-60.

Highlights

Highest reported 1,2-propanediol (1,2-PDO) titer for yeast (˃4 g L-1) from glycerol

NADH-generating glycerol catabolic pathway crucial for 1,2-PDO formation

Increased metabolic supply of precursor dihydroxyacetone phosphate essential

Strong impact of medium composition on 1,2-PDO to ethanol ratio